Reasons: Explanatory and Normative

Reasons: explanatory and normative JOSEPH RAZ University of Oxford Faculty of Law Legal Studies Research Paper Series Working Paper No 13/2007 March 2007 This paper can be downloaded without charge from the Social Science Research Network electronic library at: http://ssrn.com/abstract=999869 An index to the working papers in the University of Oxford Faculty of Law Research Paper Series is located at: <http://www.ssrn.com/link/oxford-legal-studies.html> Reasons: explanatory and normative Joseph Raz A thesis familiar by being as often disputed as defended has it that intentional action is action for a reason. The present paper contributes to the defence of a weaker version of it, namely: Acting with an intention or a purpose is acting (as things appear to one) for a reason. This thesis is weaker in two respects: (a) One would be acting intentionally if one Φs for the reason that P even if it is not the case that P or not the case that it is a reason to Φ, if one takes it to be such a reason, and (b) While all actions with a purpose or intention are intentional actions, not all intentional actions can sensibly be said to be actions done with an intention. My (automatically as it were) scratching my head, or my doodling, are intentional but not done with any intention in mind. Other intentional actions are part of sequences undertaken automatically in pursuit of some governing intention. Consider the many actions a driver performs in the course of driving one mile: accelerating, decelerating, braking, turning the wheel, operating indicators, and the like. Normally they are all intentional, but more or less automatic actions, often ones the driver is unaware of at the time. They are intentional for they are governed by an overall purpose, say getting somewhere, so that his driving is done with an intention, and therefore for a reason, whereas the many acts undertaken in the course of driving are intentional by being governed by that overall intention, rather than by being undertaken singly with any specific intention. The reasons referred to above are normative reasons. A normative reason is a fact which gives a point or a purpose to one’s action, and the action is undertaken for the sake or in pursuit of that point or purpose. Reasons, and this is the common view among writers on the subject, have a dual role here. They are both normative and explanatory. They are normative in as much as they guide decision and action, and form a basis for their evaluation. They are explanatory in that when an action for a purpose occurs the Electronic copy available at: http://ssrn.com/abstract=999869 purpose for which it is performed, the reason for the action as the agent sees things, explains its performance. In fact, I will claim, ‘reasons’ has two meanings. When the term is used in one meaning it refers to a normative consideration, when used in the other it refers to an explanatory factor. I will start by trying to make good this claim. 1. Explanatory reasons Whatever provides the answer to questions about the reason why things are as they are, become what they become, or to any other reason-why question is a Reason. Reason-why questions seek explanations and whatever provides or constitutes the explanations is the reason why whatever it explains is as it is. Needless to say I am not proposing a grammatical test. Reason-why questions can be asked without using those words. We can ask ‘what is the reason for the deformation?’, or ‘what explains the deformation?’, or use other words. What is important is the distinction between providing (or purporting to provide) information (‘It is four p.m.’, ‘She is in Sydney’) and providing (or purporting to provide) explanations. Reasons provide explanations. Some writers take propositions to provide explanations, and therefore to be reasons. As false propositions explain nothing I will join those who take facts to be explanatory reasons. One reason to take propositions (rather than facts) to be explanatory reasons is that logical and conceptual relations hold among propositions regardless of their truth. But it is well worth preserving the core idea (that reasons explain) even at the cost of occasional complexity or awkwardness of expression. Facts are reasons why; that is, they are not reasons in themselves, but reasons why something is thus and so, reasons in their function of providing an explanation. Possibly, any fact is a reason for something or other. For every fact there may be a reason-why question, in a correct reply to which it figures. To refer to a fact as an explanatory reason is to refer, at least implicitly, to a relation it has to something else: it is a reason why this or that happened, etc. Arguably, explanations are also relative to the person(s) for whom they are intended. An explanation is a good one if it explains what it sets out to explain in a way which is accessible to its addressees, that is in a way that the addressees could 2 Raz, Reasons: explanatory & normative 16/04/2007 9:40 PM Electronic copy available at: http://ssrn.com/abstract=999869 understand were they minded to do so, given who they are and what they could reasonably be expected to do in order to understand it. However, there is a clear distinction between the two relativities. No useful information is conveyed by a proposition of the form: this fact is a reason. One needs to specify something about what it explains to convey any useful information (e.g., ‘this fact explains something about the origin of life’). On the other hand, while the criteria for an explanation being a good one are relative to its addressees, what is an explanation is not. An explanation of the nature of laser radiation suitable for university students is an explanation of laser radiation, even though it is not a good explanation for primary school children. Explanatory reasons are so in virtue of their relations to what they explain, and stating that a fact is a reason is stating that it stands in the explanatory relation to what it is a reason for. As you see I am using ‘reason’ to refer to any fact which figures (nonredundantly) in an explanation, and not merely to the totality of facts all of which figure (non-redundantly) in an explanation. It is tempting to call the totality of all the facts which figure non-redundantly in an explanation a complete reason. I may occasionally use the term in order to avoid complex formulations. But if taken literally it implies more than is warranted: it implies that there is at least one comprehensive way of individuating facts, such that relative to any such scheme of individuation and object of explanation, it is either true or not, regarding each fact, that it belongs to the explanation of that object. There is reason to doubt that the explanation relation is such that it is ever true that regarding any object of explanation there is a set of explanatory facts such that it explains that object, and that adding any other fact to it is redundant so far as that explanation goes. It seems that our ways of individuating facts and the notion of explanation are such that any explanation can always be nonredundantly amplified, clarified and expanded. We should therefore take talk of complete explanation with a pinch of salt. The important point is that normally in advancing or citing reasons, non-trivial parts of ‘complete’ reasons are cited as (asserted to be) reasons, and by so citing them the speaker implicitly refers to ‘a complete reason’ (or to a disjunction of complete 3 Raz, Reasons: explanatory & normative 16/04/2007 9:40 PM Electronic copy available at: http://ssrn.com/abstract=999869 reasons), of which they are a constituent part, as the reason (for whatever they are meant to explain). We can state this point while avoiding reference to the completeness of any reason: In saying ‘that R is the reason for P’ we are saying: there is a (possibly complex) fact {R} which includes R, and which explains P. Suppose I say: the heat wave was the reason for his collapsing, and you reply: that is not so. He would not have collapsed had it been less humid. What sort of disagreement is this? You are probably pointing out that the heat does not explain his collapse by itself. It explains it only in the context of certain other facts, and it may be useful to mention some other of them (or not, as the case may be). So we do not disagree about the explanation, merely about which features of it are worth mentioning. It would have been otherwise had you said: ‘No. He collapsed because he was struck by a bullet.’ In that case we would have been advancing rival explanations. As it is we both referred to the same explanation by citing different parts of it. 1 2. Normative reasons The preceding observations explain why explanatory reasons are not much discussed by philosophers. Whatever one can say about them is better explored when studying explanations, a voluminous philosophical subject. Explanatory reasons are mostly discussed, or at least mentioned, by philosophers interested in normativity, who consider whether there is a second sense to ‘reasons’, such that in that sense ‘reasons’ refer or purport to refer to what I will call normative reasons. Is there a second sense to ‘reasons’, and if so are there such reasons? Put in different terms: are there normative reasons, and are normative reasons, if there are such, reasons independently of being explanatory reasons? Are they reasons of a different kind? This is not the same as to ask whether all reasons are explanatory reasons. I have already acknowledged that they are: it is likely that all facts, I said, can figure in some explanation or another. I will continue to assume that all reasons are facts, and 1 All these considerations apply, mutatis mutandis, to normative reasons. It is possible to try to idenfity different types of of explanatory roles for different elements in an explanation. I will not attempt such classifications. 4 Raz, Reasons: explanatory & normative 16/04/2007 9:40 PM when we refer to other things as reasons, the references can be recast as references to facts 2 , hence all reasons are explanatory reasons. That does not, however, establish the univocality of ‘reasons’. It is possible that there are facts which are reasons in a different sense while being also explanatory reasons. That they are reasons in a different sense can perhaps be established by the fact that they can explain (at least some of) what they can explain because they are reasons in a different sense of the word. I will argue that there is a second sense to ‘reasons’. When the context requires disambiguating my meaning, I will refer to reasons in this second sense as normative reasons. I will suggest that their character as normative reasons enables them to play a certain explanatory role, and thus that the way they function as explanatory reasons presupposes that they are also reasons in a different sense. It is generally agreed that the notion of a normative reason cannot be explained through an eliminative definition. That is, any explanation of it in which the word ‘reason’ does not occur will include another term or phrase whose meaning is close to that of ‘a reason’ so that those who puzzle over the nature of reasons will not be helped by the definition. It will raise similar puzzles in their mind. We explain the notion of a normative reason by setting out its complex inter-relations to other concepts. Not to explain, but to minimally locate what we are talking about, we can say that normative reasons, if there are such, count in favour of that for which they are reasons. They have the potential to (that is, they may) justify and require that which they favour. Those who wish to deny that normative reasons are a distinct kind of reasons may claim that normative reasons are simply explanatory reasons that differ from others in providing explanations of a special kind of facts. After all explanatory reasons are often classified by what they explain: individual events, or laws of nature; motivations or pains, etc. The distinctness of the object of explanation does not 2 I do not mean to suggest that we do or should refer only to facts as reasons. I follow this usage for convenience’ sake only, but by the same token not only facts can be taken to be explanatory reasons. For example, I refer to hypnosis as a reason in the text below. 5 Raz, Reasons: explanatory & normative 16/04/2007 9:40 PM require different senses of ‘reasons’. Reasons are just the facts which explain. Different kinds of facts may explain different kinds of phenomena but they explain them, and are reasons, in the same sense of ‘explanation’ and of ‘reasons’. If so-called normative reasons can favour, justify or require, is it not simply a way of saying that they are the facts which explain why it is a fact that something is favoured, justified or required? That is they are facts explaining why the proposition that this or that is favoured, justified or required is true. If so does it not follow that there is no different sense of reason here, only a different object of explanation? But normative reasons do not always justify or require what they favour. Nor is it always the case that what they favour is (non-relationally) favoured. That depends on what else is true of it. When they neither justify nor require it, they cannot explain why it is justified or required. And when what they favour is not (non-relationally) favoured they cannot explain why it is favoured. Can we do better by modifying the suggestion: would not those reasons explain why what they favour is favoured by them? Only if one thinks that a fact can explain itself; or if you like, only in the pickwickian sense in which that A favours B explains why B is favoured by A. Clearly here talking of explanation is otiose. In any case it does not dispose of the claim that normative reasons do something other than explain. Their explanatory use is secondary, and depends on the fact that they favour what they favour, a fact which sets them apart from other explanatory reasons. The existence of a normative relation: that one thing is a reason for another, is, on this suggestion, the object of the explanation. But for there to be something to explain there must be normative reasons, that is reasons in a sense which is independent of the explanatory sense of ‘reasons’. Or consider another tack: if reasons for belief are reasons only in being explanatory reasons, what do they explain? The answer may appear obvious: they explain why the person who has the belief, has it. They explain his believing. If that is meant as a universal truth about explanations of believing, that is of why people who have a belief believe as they do, then this is false. For example, the reason Jamie believes a certain proposition may be neither the reasons there are for his belief, nor 6 Raz, Reasons: explanatory & normative 16/04/2007 9:40 PM the reasons for it of whose existence he is or can become aware. The reason for his believing may be that he was hypnotised to believe it. But hypnotic suggestions are not reasons for belief. So while the reason for his believing may be that he was hypnotised that is not a reason for the belief, it is not a normative reason for believing that proposition. Nor is it necessarily the case that the reasons to believe a proposition are the facts which explain that belief (that believing) if the belief is rational or rationally held. The reason which explains the believing looks back to its causes (the causes of having it or of still having it). The rationality of believing depends on one’s openness to critical evaluation of the belief, one’s ability and willingness to revise or reject it were the evidence to point that way. One may say that reasons for a belief are those facts which explain the believing, the acquisition of the belief, when it was rationally induced. But this view allows that ‘reasons’ is ambiguous between explanatory reasons, which, presumably, can explain all beliefs, and normative reasons, e.g. reasons for belief, which also explain those beliefs which were rationally arrived at or are rationally sustained, that is beliefs arrived at or sustained because of reasons for the beliefs. Regarding the latter kind of reasons their ability to explain the believing depends on the fact that they are normative reasons, reasons which can justify a belief, whether or not they also explain it, and which explain beliefs as rational or justified because they are normative reasons. Can one avoid this conclusion by claiming that reasons for belief explain why one ought to believe? When one ought to believe something, reasons for believing it, reasons which justify believing it, will be among the considerations which explain why one ought to have that belief. That is because one ought only to have beliefs which one is warranted, justified, in having. But that explanatory role depends on and presupposes the distinctive normative role of normative reasons. Moreover, reasons for belief do more than justify belief which one ought to have. Not infrequently the evidence for a proposition is such that while believing it is warranted, failure to believe it is not irrational. In such cases it is false that one ought 7 Raz, Reasons: explanatory & normative 16/04/2007 9:40 PM to believe in the proposition, yet one has reasons to believe in it, reasons which are sufficient to justify or warrant the belief. On other occasions, there are reasons for belief which nevertheless do not warrant belief. They are too weak, or the person concerned has not inquired enough to warrant his reaching any conclusions. In such cases having the belief is usually irrational. In such cases it is difficult to find anything that the existing or available reasons for belief explain 3 (other than that there are reasons for that belief), unless someone happens irrationally to adopt the belief because of those inadequate reasons. It is relevant here that we regularly refer to reasons for belief independently of any explanatory context, i.e. when reasoning about what to believe, which is not the same as reasoning about what would explain the belief once we have it (and remember that – for reasons given - reasoning what to believe is not to be confused with reasoning about what one ought to believe). 3. Normative reasons and ought-propositions – Broome’s reasons: I take reasons to be the key to an understanding of normativity. Possibly one or several other concepts can play a similar role, though I suspect that they will not be among the concepts normally used by English speakers today. This section explores the suggestion, as defended by Broome, that the basic normative concept is that of ought-facts, and that reasons are to be explained by their role in explaining ought-facts. I will continue to use the more common terminology of true ought-propositions, meaning true propositions which can be expressed in sentences containing an ought operator (used in their primary meaning). My discussion of Broome’s views has two objects: first, to understand why ought is not the basic concept; second, to explain why Broome’s understanding of reasons is partial and misleading. Broome’s view revolves around three theses: 3 To explain qua reasons for belief. They may explain any number of things which have nothing to do with their force as reasons for belief. They may, e.g., explain why it is raining. 8 Raz, Reasons: explanatory & normative 16/04/2007 9:40 PM A) Some reasons are perfect reasons: ‘A perfect reason for you to Φ is … a fact that explains why you ought to Φ’. 4 B) Other reasons are pro tanto reasons: ‘A pro tanto reason for you to Φ is a fact that plays the for-Φ role in a potential or actual weighing explanation of why you ought to Φ, or in a potential or actual weighing explanation of why it is not the case that you ought to Φ and not the case that you ought not to Φ’. (41) C) Some true ought-propositions cannot be explained by normative reasons. The first thing to note about the theses is that they are consistent with the view that ‘reason’ is ambiguous between the explanatory and the normative sense. Arguably only facts which constitute normative reasons can explain true oughtpropositions. Broome repeatedly asserts (34-5) that perfect reasons are not normative; they merely explain a normative fact. But there is no reason to think that normative reasons cannot explain other normative facts, and, as I point out below, it is difficult to understand his pro tanto reasons except as normative facts, a species of normative reasons, which, among other things, play a part in explanatory reasons of the kind he calls ‘perfect reasons’, namely in the explanation of the truth of some ought propositions. So “perfect reasons” can be normative. Broome is right, however, to point out that all perfect reasons, i.e. all explanations of why one ought to do this or that or to believe this or that, will include elements which are not normative reasons. They all require a closure proposition, that is a proposition which states that the explanation includes all the factors relevant to the truth of the ought-proposition, and closure propositions are not themselves statements of reasons. So, if, as Broome stipulates, “perfect reasons” are simply explanatory reasons, then – and this is my view – they include some normative reasons, and some other elements. Furthermore, my view of normative reasons is more than consistent with the claim that normative reasons explain the truth of true ought-propositions; given what ‘ought’ means, my view entails that true ought-propositions are to be established by 4 Broome (2004) 35 9 Raz, Reasons: explanatory & normative 16/04/2007 9:40 PM normative reasons. Before coming to that let me note four points of disagreement between Broome’s view and mine. First, according to Broome some true ought-propositions cannot be explained by normative reasons. Second, his view implies that all normative reasons are pro tanto reasons, as he understands them. Third, furthermore his view implies that normative reasons are mere explanations of why ought-propositions are or are not true. Their normativity, as it were, exhausts itself in being such explanations. Beyond these looms a fourth disagreement about the content of ought-propositions. Broome offers few examples of explanations of ought-propositions which do not include normative reasons (pp. 43-47). They all depend on his characterisation of normative reasons as coming with weights. For example, he writes: You ought not to believe that it is Sunday and that it is Wednesday. A plausible explanation of why not is that ‘It is Sunday’ and ‘It is Wednesday’ are contrary propositions and you ought never to believe both a proposition and its contrary. (42-3) He rightly points out that no weights are involved in this piece of reasoning. But if it were true (which it is not) 5 that you ought never to believe both a proposition and its contrary, then there is a reason for not believing that it is Sunday and that it is Wednesday. Whatever establishes that you ought not have contrary beliefs is also a reason for not having these two beliefs. Once we use ‘reasons’ in the normal way, to enable us to refer to all normative reasons, the counter examples disappear. So, Broome’s claim that some ought propositions cannot be explained by normative reasons rests on his narrow conception of normative reasons as “pro tanto reasons”, as he understands them. His explanation of how pro tanto reasons explain the ought-propositions which they explain is very narrow, possibly applying to no reasons at all, with the result that many reasons are ignored by him. He stipulates that pro tanto reasons contribute to weighting explanations: … there are reasons for you to Φ and reasons for you not to Φ. Each reason is associated with a number that represents its weight. The numbers associated with 5 For reasons I explored in ‘The Myth of Instrumental Rationality’. 10 Raz, Reasons: explanatory & normative 16/04/2007 9:40 PM the reasons to Φ add up to more than the numbers associated with the reasons not to Φ. That is why you ought to Φ (36-37). He relaxes these conditions: weight need not be exact and the function of total weight from component weights need not be additive. Broome’s explanation of pro tanto reasons combines two elements. First, these reasons come with weights, and what one ought to do is some function of those weights. Second, their being reasons for something entails the possibility of the existence of a reason against that very thing. I am not aware of any ought-proposition which can be explained by reference to weights in the way suggested by Broome. Be that as it may, it seems plain that weights play no role in our understanding of many reasons for actions which are not conclusive ones, but which determine what to do, to the extent that that is determined by reasons. The very notion of an associated weight is hard to make sense of. Broome does not offer any justification for the weight-related view of pro tanto reasons, unless one takes his reference to the use of weighing and balancing metaphors by various writers as a justification. But metaphors are exactly that. I regularly write about one reason defeating another. But that is hardly a reason to attribute to me the view that conflicting reasons are opposing fighters, who engage in some form of combat. I have suggested that the non-metaphorical point is that in deliberating about what one ought to do propositions about various reasons are relevant, and it is by reasoning from them, and about their interrelations and so on, that the conclusion is drawn. 6 In fact, that aspect of Broome’s view is not essential to the thesis that reasons are just explanatory facts, and I will ignore it. The second element in the definition of pro tanto reasons is that they are normative reasons which are necessarily not conclusive. Broome rightly points to a difference in the logical standing of (what I regard as) different normative reasons. Some are such that there cannot be any reasons which conflict with them. Others, his ‘pro tanto reasons’, do not have that feature, that is their being a reason for something is consistent with there being a reason against that very thing. The contrast is real and 6 Raz (1975) 11 Raz, Reasons: explanatory & normative 16/04/2007 9:40 PM significant. But so are many other distinctions between types of reasons. Arguably, the difference between epistemic and practical reasons is more fundamental, but both are normative reasons. 7 There is no reason to think that normative reasons are confined to those which manifest either of the features by which Broome defines pro tanto reasons. We can now return to the third of Broome’s three contested claims, that is that normative reasons merely contribute to the explanation of ought-propositions. I suggested already in the previous section that that is not all that they do. I will illustrate the point with two examples: In cases of akrasia people act for what they believe to be the lesser reason. For akrasia to be possible it must be possible that they are right in that belief. 8 So imagine cases in which they are right. I knowingly act for a lesser reason. I am not acting as I ought to act, and I know it. But I am acting for a (genuine) reason. It is merely one which, as I am aware, is defeated in the circumstances. The reason for which I act is a normative reason, and it explains my action. Its being a normative reason and its success in explaining my action, which is due to its ability to motivate me, as normative reasons can, do not depend on its contribution to the explanation of any ought-proposition, not even to the explanation of the falsity of any such proposition. This shows that, even though normative reasons may contribute to the explanation of some true ought-propositions, their relevance in cases of akrasia goes beyond any such contribution. The second example is drawn from a case recognised by Broome. He mentions that sometimes it is not the case that one ought to Φ, nor is it the case that one ought not to Φ. As he says such cases may belong to different subcategories (38-9). In some there are no normative reasons either for or against Φ-ing. In others there are reasons pro and con Φ-ing which do not defeat one another. Now here the existence of normative reasons is essential to elucidate the difference between these two types of case, though there is no difference between them regarding which ought-propositions 7 See my ‘Reasons: Practical and Epistemic’ 8 12 Raz, Reasons: explanatory & normative 16/04/2007 9:40 PM apply to them. Again, we see that normative reasons do more than explain oughtpropositions. These reflections point to the nature of the relations between reasons and ought-propositions, which I will sketch in rough metaphorical terms. Normative reasons are facts which have normative bearing or force. They are called ‘reasons’ because they can serve as stepping stones in reasoning about what to believe or what to do. Deliberating from the reasons which apply to us we become aware of the attractions and drawbacks of options. We may reach a variety of conclusions: that we have a duty to do something (and we may have duties which we ought not to fulfil); that we have a right to do something; that certain options are acceptable (we may have a right to do unacceptable things); that it would be prudent to take some actions; that we should take some actions; that it would be irrational not to take them; and others. One such conclusion, different from any of the ones I mentioned, is that we ought to take some action. Ought-propositions are not the centre of practical thought. Nor are they the foundations on the basis of which we can understand reasons. Rather, they are one of a variety of propositions whose truth conditions are the existence or absence of some normative reasons or others. There is one particular factor which complicates the explanations of oughtpropositions. Let us focus on propositions of the form: ‘When C, P ought to Φ’ (where ‘C’ stands for circumstances, ‘P’ for a person, or a set of people, and ‘Φ’ for an action, a doing or an omission). When such propositions are very specific they state what one has conclusive reason to do. For example: that John Doe ought to give Jane Roe £5 by midnight is true only if there is a conclusive reason for John Doe to give Jane Roe £5 by midnight. But this is not true of general ought-propositions. This is particularly clear regarding universal ought-propositions. Here are a few examples: ‘People ought to pay their debts punctually’; ‘People ought to be kind to their grandparents’; ‘One ought to vote in parliamentary elections’. Those who believe these and similar propositions do not necessarily believe that there always are conclusive reasons to be kind to one’s grandparents, to repay debts punctually or to vote in parliamentary elections. At the 13 Raz, Reasons: explanatory & normative 16/04/2007 9:40 PM very least it should be clear that compliance with these three ought-propositions may on occasion conflict: On a particular day it may be the case that one would not be able to vote in the election if one were to pay one’s debt punctually and vice versa. There are two kinds of universal ought-propositions which are true only if there are conclusive reasons to behave as they indicate: First, there are conceptually true propositions of that kind. If murder is unjustified intentional homicide then ‘One ought not to murder’ is true because there is always a conclusive reason not to murder, and, on our assumption, that is a conceptually necessary truth. Second, there may be some so-called moral absolutes (or absolute reasons of other kinds) namely reasons (or combinations of reasons) which defeat all possible (combinations of) conflicting reasons. For example, possibly it is true that ‘In all states the law ought to prohibit torture, without any qualification’, meaning among other things that all governments ought to see to it that their law prohibits torture. But most universal ought-propositions do not belong to either kind. Are they all false? One cannot respond that they are implicitly relative. ‘One ought to vote in general elections’ and ‘so far as one’s civic duties are concerned one ought to vote in general elections’ are distinct propositions, and in any case even the relativised proposition could be true even though it is not the case that in so far as one’s civic duties only are concerned one always has a conclusive reasons to vote in general elections. Sometimes one’s civic duties require abstaining. The same will be true of any non-conceptually necessary relativisation. Nor can one maintain that universal ought-propositions are abbreviations of very detailed propositions which include a complete list of exceptions, all the circumstances in which one does not have a conclusive reason to act as the proposition indicates. People who have complete knowledge and understanding of what they believe when they believe a universal ought-proposition do not necessarily know all these exceptions. In fact it is plausible to think that no one can know all the exceptions (i.e. that no one can know either a list of them or a generalisation stating them, which is not true merely for logical or conceptual grounds), while one can have complete knowledge and understanding of (the content of) ought-propositions. Hence 14 Raz, Reasons: explanatory & normative 16/04/2007 9:40 PM the propositions normally expressed by universal ought-propositions are what they appear to be. They are not (identical with) detailed and exceptionless propositions. It is one of the virtues of the concept of normative reasons that it enables us to think about normatively complex and indefinitely changeable situations, helping us to marshal their normatively significant features into forms which facilitate coherent deliberation. So how are universal ought-propositions related to specific ones and to reasons? I think that propositions of the form ‘When C, P ought to Φ’ are true just in case, and because, there is a reason (or a number of reasons) which applies whenever C is the case, and which in at least some instances of C is a conclusive reason for P to Φ. 9 On this view it is a conceptual truth that there are normative reasons which explain why one ought to Φ, when one ought to. This account of the truth conditions of practical ought-propositions 10 specifies the same truth conditions for specific and for universal ought-propositions. It is merely that as specific propositions apply only to one occasion they are true only if on that occasion there is a conclusive reason to do as they indicate. This account cannot be taken to be a general account of the meaning of oughtpropositions. For one thing it does not generalise to epistemic oughts, to what we ought to believe. 11 It has other limitations as well. It is too simple to capture the nuanced ways in which ‘ought’ is standardly used, and therefore also to account for the nuanced differences among propositions in whose expression it features. It does explain, however, why practical ought-propositions cannot play any foundational role in understanding practical thought. 9 10 11 Note that I am referring to simple unqualified ought-propositions (displaying the general form ‘X ought to Φ’ or ‘when C X ought to Φ). Their meaning varies when qualified: one always ought to Φ may mean that one’s reason for Φ-ing is always conclusive, etc. First suggested by me in [1978]. I suspect that ‘P ought to believe that ....’ indicates that it would be irrational (or more weakly, a failure of rationality) for P not to believe that ... . one may well reject a true proposition that one ought to Φ, or just fail to Φ, without committing any rational fault. 15 Raz, Reasons: explanatory & normative 16/04/2007 9:40 PM 4. The Normative/Explanatory Nexus It seems plausible to assume that reasons in both senses are called ‘reasons’ because of their connection to Reason. But there is a closer connection between them which explains the common name. Briefly said it is that normative reasons provide the standard explanations of beliefs and of actions done with an intention or a purpose. Moreover, it is a necessary condition of any fact being a reason that, when conditions are appropriate, it provides such an explanation. Put another way, epistemic reasons can explain (or figure in an explanation of) beliefs, and practical reasons can explain (or figure in an explanation of) actions performed with an intention or purpose. This point is generally recognised, though sometimes neglected. It expresses the thought that normative reasons can guide agents, that is that they can move agents, who are aware of them, to action, belief and the like. Hence they can feature in explanations of such actions, beliefs and the like. In further exploring that idea I will not be looking for a characterisation of the causal or other mechanisms on the existence of which these explanations depend. I will merely try to characterise the kind of explanation involved. We can start the exploration in the company of Bernard Williams, since the point was crucial to his argument for reason internalism. 12 How do normative reasons explain and what do they explain? Following Williams I will explore this regarding practical reasons only. Similar considerations apply to epistemic reasons. Obviously, reasons for an action do not always explain the action, even when it was performed. It may have happened accidentally, and even when intentional, the intention may have been motivated by something else, either by some other reason for that action or by a mistaken belief that there is some other reason. So the point is not that whenever one does what there is a reason to do one acts for that reason. Nor is it that there are no other, non-reason-related, explanations for an action (hypnosis, statistical explanations and others). Rather the point is that normative 12 I will not consider the merits of any form of internalism or externalism about reasons, nor Williams’s own argument for internalism. Like some other writers I think that the contrast is more confusing than helpful. My own view will be clear enough. Its classification as a form of internalism or externalism is immaterial. 16 Raz, Reasons: explanatory & normative 16/04/2007 9:40 PM reasons must be capable of providing an explanation of an action: If that R is a reason to Φ then it must be possible that people Φ for the reason that R and when they do, that explains their action. Or, as Williams puts the point: If there are reasons for action, it must be that people sometimes act for those reasons, and if they do their reasons must figure in some correct explanation of their action. 13 Furthermore, the role reasons play in the explanation must be of a certain form. If that R is a reason to Φ, then it must be possible that awareness that R motivated the agent to Φ. Sometimes the phrase ‘motivating reasons’ is invoked in such contexts. I will not use it myself, for it is liable to confuse. Sometimes the phrase is used to refer to a kind of explanatory reasons for actions, those which explain them by explaining that they were motivated by belief in the existence of a (normative) reason. That sense is much narrower than the natural understanding of the phrase (motivating reasons being reasons explaining actions by their motivations) on the one hand, and is wider than the way the phrase is sometimes used (to refer to reasons explaining actions by reference to being motivated by awareness of reasons for them) on the other hand.14 [] Back to business: what matters for our purpose is not that facts that are normative reasons can explain (that they can figure in the explanation of) actions. Just about any fact can (given appropriate circumstances) figure in some explanation of some actions. The normative/explanatory nexus requires that the potential explanatory role of facts which are normative reasons depends on and presupposes their normative force: it has to be that they can explain because they are normative reasons. That I promised can explain my promise-keeping action of giving a copy of my 13 14 Bernard William, (1975) 102 See Jonathan Dancy 2000, 6: ‘I have characterised the distinction between the reasons why we do things and the reasons in favour of doing them in terms of the motivating and normative. In doing so I have tried to avoid any suggestion that we are dealing here with two sorts of reasons. … the same reason can be both motivating and normative. A reason for acting can be the reason why one acted’. I agree. The same fact can be both a normative reason for action and an explanatory reason of why one acted. 17 Raz, Reasons: explanatory & normative 16/04/2007 9:40 PM book in a way in which the impact of low atmospheric pressure cannot. No doubt being in the mountains where atmospheric pressure is low can explain some aspect of my conduct, but the explanation (assuming the normal scenario) is of a completely different kind. What is this special way in which normative facts figure in explanation? I take my promise as a reason for the action, recognise it as such even before the action, and take the action to be sensible because it is supported, perhaps even required, by my promise. The low pressure just affects me, independently of whether I am aware of it, and regardless of what I think of it. We can baptise this thought ‘normative/explanatory nexus’, namely that regarding every normative reason, it is possible for it to feature in an explanation of the action for which it is a reason as a fact whose recognition motivated the agent to perform it, and guided him in its performance. This last condition, that awareness of the reason guides the performance, and does not merely initiate it, is taken for granted in common reason-explanations of actions, but is sometimes ignored to their detriment by more formal philosophical accounts. Harry Frankfurt pointed out its importance, and Kieran Setiya suggested that it solves the problem of deviant causation (so far as basic actions are concerned) which afflicts accounts of intentional action such as Davidson’s.15 Williams explains an important aspect of this nexus. To see it we need to remind ourselves how Williams’s view differs from that of some other Humeans. First, Williams finds no problem in the fact that reasons may be facts about how things are in the world and not only about the agents’ beliefs and desires. Of course in that case they could figure in explanations of the agents’ conduct only if those agents are aware of them. But that is consistent with the normative/explanatory nexus as he (and I) understand it. Furthermore, Williams does not object to the thought that beliefs can motivate. He does not endorse the view that only desires can be reasons, since only desires motivate, and therefore only they can explain actions in the right way. This is neither true, nor is it true to Williams’s account. He allows, for example that 15 See Frankfurt ‘The Problem of Action’ in THE IMPORTANCE OF WHAT WE CARE ABOUT pp. 69, 72ff, and K. Setiya, REASONS WITHOUT RATIONALISM, 31-32. 18 Raz, Reasons: explanatory & normative 16/04/2007 9:40 PM ‘there are some cases of an agent’s Φ-ing because he believes that there is a reason for him to Φ, while he does not have any belief about what that reason is.’ (107) More generally he asks: ‘Does believing that a particular consideration is a reason to act in a particular way provide, or indeed constitute, a motivation to act? … Let us grant that it does – this claim indeed seems plausible …’ (ibid.). The crux for Williams is not directly in the possibility of being motivated by one’s beliefs but in the way one could acquire such motivating beliefs: the basic point lies in recognising that the external reasons theorist must conceive in a special way the connexion between acquiring a motivation and coming to believe the reason statement. For of course there are various means by which the agent could come to have the motivation and also to believe the reason statement, but which are the wrong kind of means to interest the external reasons theorist. Owen [Wingrave – in James’s story] might be so persuaded by his father’s moving rhetoric that he acquired both the motivation and the belief. But this excludes an element which the external reasons theorist essentially wants, that the agent should acquire the motivation because he comes to believe the reason statement, and that he should do the latter, moreover, because in some way he is considering the matter aright. (108-9) This is indeed a required element. The initial thought that normative reasons must be capable of explaining is not that (taking the matter beyond Williams’ example, but in the spirit of his remarks) one could accidentally come to be motivated by awareness of the reason. Awareness of the reason must non-accidentally motivate, and it must motivate, as Williams puts it, because the agent ‘is considering the matter aright’. But what is that way? Here we have to go beyond Williams, though without conflicting with what he says in the quoted passages. Williams’s phrase ‘considering … aright’ suggests, first, that the explanation relates to rational agents, and depends on their exercise of their rational powers, and therefore (given the implausibility that the motivation can be generated in the right way by some external circumstances 19 Raz, Reasons: explanatory & normative 16/04/2007 9:40 PM surrounding the believing) second, that the way that the belief has to explain the motivation is by having the content it has, by what it is a belief about. This leads in several steps to the requirement that the reason itself figure in the explanation. First, the belief must be belief in the fact which is a reason, and include realisation of its character as a reason. Motivations to perform a particular act would not be reliably and rationally brought about (or constituted) by a belief unless the belief was a belief about a reason to perform that act. Second, the belief itself must be explained by the existence of the reason, and it must be acquired or maintained in a rational way. Therefore, third, the reason itself figures in the explanation of the action, in being part of the explanation of the belief which motivated the action. And finally, fourth, the belief, the awareness of the reason (where it is rational and true) must not only prompt, but guide the action. A typical objection relies on the fact that, as was pointed out at the beginning of the article, an act is intentional, and done with a purpose even if the belief which motivated it is false. It follows, goes the objection, that the fact which renders the belief true (when it is true) cannot be part of the explanation of the action. It has, of course, to be admitted that when the belief is false (a) the action can be explained, and (b) its explanation as intentional must include reference to the belief that there was a reason for it (as was argued in the first point above), and (c) as there is no reason to make that belief true, no reason can be part of the explanation. But that applies to the explanation of intentional actions based on false beliefs. In such cases the explanation does not refer to a normative reason for the action. Even if there were such reasons they do not explain why the action was performed. However, the objection continues, if the belief alone is sufficient to explain intentional actions when it is false it must also be sufficient to explain actions when the belief is true. The further factor, the existence of the reason is not necessary to the explanation which is, as is shown by the case of false beliefs, adequate without it. It has to be admitted that citing the belief, without adding that it is rational and true does explain the action, and shows it to be intentional. Furthermore, it is plausible to think that ‘being an explanation of’ is not a transitive relation. Sometimes even if C 20 Raz, Reasons: explanatory & normative 16/04/2007 9:40 PM explains B and B explains A, C does not explain A. So even if the existence of the reason explains awareness of it as a reason, we need something additional to show that it can also explain action for that reason. But that is consistent with the possibility that an explanation which includes the reason among the explanatory factors is a better explanation of the intentional actions to which it applies. To be sure the reason is not part of the explanation of the action just by being an element in the explanation of the belief which prompts the action. It has to play a role in the explanation of the action itself, especially in its explanation as intentional. The practice of explanation shows that in fact the reason does figure in explanations of actions: Why did I go to Chamonix for my holidays? Because it is so beautiful there; why am I rushing to my office? Because I promised to meet a student there in ten minutes’ time, etc. It is, however, one thing to know that reasons can figure in explanations of action, it is another to understand why this is so, that is, why they explain not only belief in their existence, but also the actions this belief leads to. Reasons are part of the explanations of rationally held beliefs because they explain that, and in what way, they are rational. Beliefs and what they are about are related contingently, but not accidentally. Their intimate relationship is expressed by the fact that beliefs are defective if we hold them not because things are as they say they are. A healthy belief, one may say, a belief which is as beliefs should be, is a belief which one rationally has because things are as it says they are. That may explain why what explains the belief also explains what we do when we rationally react to it. 16 After all explanations which end with the belief, and do not refer to its rationality or to the reason itself, cannot explain whether the intentionality was successful or a failure. An intentional action which cannot be explained by the reason which motivated it is one which fails in its own terms. When acting for a purpose we aim to do something for an adequate reason. When the reason we intended to follow was not there to follow, the action turns out to be something other 16 Though even if we believe something because things are as we believe (say I believe that there is a tree in my garden because there is one) what explains the belief need not explain an irrational reaction to it (e.g. if I react to my belief by irrational panic that the tree will fall on me and will kill me on my birthday). That belongs to the fourth and last stage in the account. 21 Raz, Reasons: explanatory & normative 16/04/2007 9:40 PM than it set out to be. It was, to that extent, a failed intentional action. In explaining intentional actions we need to make clear what they were meant to be and whether they succeeded, and that means that where available, the reason is part of the explanation of the action as a successful intentional action. Finally, to the fourth claim, that awareness of the reason must guide the action. I do not mean that otherwise the motivation would be irrational. There are many possible causal routes from a belief in a reason to motivation, which while not irrational, are adventitious. If normative reasons are to meet a meaningful explanatory potential requirement they must be capable of explaining through belief in their existence qua reasons. The element of guidance can be understood by analogy to a negative feedback mechanism: we, automatically and normally without being conscious of the fact, monitor the performance of the intentional action such that if it deviates from the course we implicitly take to lead to its successful completion we correct the performance, bringing it back to the correct path, or interrupt it, when we fail to correct it. So, the claim is that one’s action is guided by a reason just in case that one is motivated by the reason, through awareness of it, in a way which is manifested by the (normally unconscious) self-correcting process of tracking the success of the process of performing the action. Another possible worry may be generated by my claim that reasons explain actions through the mediation of awareness of the reasons as reasons. It may be thought that this implies possessing concepts and having beliefs which many people who act for reasons do not have. This worry, if justified, may not disprove the letter of the normative/explanatory nexus. After all it requires nothing more than the possibility of certain explanations. But it would go against its spirit. For surely, the nexus is meant to relate to a standard form of explanation which can be used to explain any action for a reason. The worry is, however, unjustified. Having concepts and conceptual thought involves recognising the implications of the concepts we use, and being guided by them, reacting to them appropriately and trying to adjust our responses when becoming aware that they are inappropriate. The inferential connections between concepts and between thoughts can be spelt out in 22 Raz, Reasons: explanatory & normative 16/04/2007 9:40 PM general terms. We may say that there are principles which spell out these connections. Knowledge of such principles, however, requires having concepts which not all those who have the thoughts that they explain need have. For example it may require having the concept of a reason, which not everyone capable of conceptual thought has. This means that the gap between having concepts and knowing how to employ them, on the one hand, and being able to think about the principles which govern one’s understanding of those concepts, on the other, is greater than may appear. I can treat a promise as a reason, and as we may say, implicitly know that it is a reason, without being able to understand any statement of the principle which sets out what I understand when I understand the notion of a promise. We are here in territory which was explored by Brandom, and using his terminology we may say that making things explicit is more difficult than is sometimes thought, for it may require concepts one need not have to have the implicit knowledge. I therefore conclude that it is justified to say that the normative/explanatory nexus does not require excessive conceptual mastery, nor excessive conceptual knowledge. As was already explained, the requirement that normative reasons explain through agents being aware of reasons as reasons is necessary to ensure that we refer to the right kind of explanation, and the right kind of explanation is explanation mediated by our rational faculties. A final possible dissatisfaction I will mention here is that none of the above constitutes an explanation of the productive process which leads to awareness of the reason and from there to the motivation and the action. That is, of course, true, but I doubt that it is a drawback. Welcome as such explanations are, they are not needed for an understanding of the normative/explanatory nexus. Indeed so far as that goes they may do too much. As stated the nexus is sufficient, for those who understand it, to distinguish cases in which an action is done for a normative reason (and can be explained relying on the nexus) from other cases. It can be rightly pointed out that the ability to distinguish those cases made possible by the statement of the nexus is not sharp, that it leaves us undecided in many cases in ways which no further explanation can resolve. But that is just how things should be. The phenomena are not sharp 23 Raz, Reasons: explanatory & normative 16/04/2007 9:40 PM because they are defined by our concepts, and our concepts are vague, leaving the phenomena they apply to vague. So, the normative/explanatory nexus states that necessarily normative reasons can explain the actions, beliefs and the like of rational agents. The relevant explanations explain agents’ beliefs or actions, etc. as a result of their exercise of their rational powers, leading to awareness of the facts which are reasons qua reasons, and to rational reaction to this awareness. 24 Raz, Reasons: explanatory & normative 16/04/2007 9:40 PM